feat: support history, norms and goals for LLM

ref: N25B-299

src/control_backend/agents/bdi/bdi_core_agent/bdi_core_agent.py

@@ -12,8 +12,11 @@ from control_backend.agents.base import BaseAgent
 from control_backend.core.agent_system import InternalMessage
 from control_backend.core.config import settings
 from control_backend.schemas.belief_message import Belief, BeliefMessage
+from control_backend.schemas.llm_prompt_message import LLMPromptMessage
 from control_backend.schemas.ri_message import SpeechCommand
 
+DELIMITER = ";\n"  # TODO: temporary until we support lists in AgentSpeak
+
 
 class BDICoreAgent(BaseAgent):
     bdi_agent: agentspeak.runtime.Agent
@@ -112,7 +115,9 @@ class BDICoreAgent(BaseAgent):
             self._add_belief(belief.name, belief.arguments)
 
     def _add_belief(self, name: str, args: Iterable[str] = []):
-        new_args = (agentspeak.Literal(arg) for arg in args)
+        # new_args = (agentspeak.Literal(arg) for arg in args)  # TODO: Eventually support multiple
+        merged_args = DELIMITER.join(arg for arg in args)
+        new_args = (agentspeak.Literal(merged_args),)
         term = agentspeak.Literal(name, new_args)
 
         self.bdi_agent.call(
@@ -178,21 +183,37 @@ class BDICoreAgent(BaseAgent):
         the function expects (which will be located in `term.args`).
         """
 
-        @self.actions.add(".reply", 1)
-        def _reply(agent, term, intention):
+        @self.actions.add(".reply", 3)
+        def _reply(agent: "BDICoreAgent", term, intention):
             """
-            Sends text to the LLM.
+            Sends text to the LLM (AgentSpeak action).
+            Example: .reply("Hello LLM!", "Some norm", "Some goal")
             """
             message_text = agentspeak.grounded(term.args[0], intention.scope)
+            norms = agentspeak.grounded(term.args[1], intention.scope)
+            goals = agentspeak.grounded(term.args[2], intention.scope)
 
-            asyncio.create_task(self._send_to_llm(str(message_text)))
+            self.logger.debug("Norms: %s", norms)
+            self.logger.debug("Goals: %s", goals)
+            self.logger.debug("User text: %s", message_text)
+
+            asyncio.create_task(self._send_to_llm(str(message_text), str(norms), str(goals)))
             yield
 
-    async def _send_to_llm(self, text: str):
+    async def _send_to_llm(self, text: str, norms: str = None, goals: str = None):
         """
         Sends a text query to the LLM agent asynchronously.
         """
-        msg = InternalMessage(to=settings.agent_settings.llm_name, sender=self.name, body=text)
+        prompt = LLMPromptMessage(
+            text=text,
+            norms=norms.split("\n") if norms else [],
+            goals=goals.split("\n") if norms else [],
+        )
+        msg = InternalMessage(
+            to=settings.agent_settings.llm_name,
+            sender=self.name,
+            body=prompt.model_dump_json(),
+        )
         await self.send(msg)
         self.logger.info("Message sent to LLM agent: %s", text)
 

src/control_backend/agents/bdi/bdi_core_agent/rules.asl

@@ -1,3 +1,6 @@
-+user_said(Message) <-
+norms("").
+goals("").
+
++user_said(Message) : norms(Norms) & goals(Goals) <-
   -user_said(Message);
-  .reply(Message).
+  .reply(Message, Norms, Goals).